A new study proves that high sample rates really DO sound better. Or does it ? As usual it's a little more complicated than that... in this episode we investigate why, and also discuss how to test this stuff for yourself, plus - when you might choose NOT to mix at Abbey Road, why all CD players don't sound the same, and why Ian hates the sound of swirling ultrasonic birdies. And, what that actually means!
Links
AES Paper: A Meta-Analysis of High Resolution Audio Perceptual Evaluation
Test files for intermodulation distortion
The importance of loudness matching
Why mp3 sucks - and how to hear it
This was a great episode and the "nerdy" talk was indeed fascinating. Keep up the great show
Michael Johnson - "Then and Now"
Hi Ian, of all Albums, CD's, tapes etc I've heard this one towers above them all. When I reference it against other recordings it stands out a mile, even in MP3! A superb recording/mastering effort, please let me know what you think.
Croaky
Ultrasonic swirling Birdies! I will never be able to hear this without laughing . Love it! I am thinking Ian should start a band with this name!
great job as usual.
cheers
• There are people who listen to sound and people who listen to music... 😉
A good old distorted 7-inch single by The Who is certainly more enjoyable to me than any perfectly mastered HD audio by... well, I won't point anyone out. 😀
Léo Ferré, an old French singer, said in a song that people who count feet are not poets but surveyors (I hope the translation is correct), the same could apply to many audiophiles, counting bits and Hz instead of simply enjoying music... 🙂 — but it's their absolute right, after all.
• I think you're pretty generous saying people can't generally hear above 17-18 kHz (if I recall well): I bet the vast majority out there can't hear anything above 13 kHz, after a certain age (30? 40?), and that will be even worse with the mp3 players generation.
• Though, at the recording/mixing stage, it can make sense using higher resolutions IF you know that you'll have to time-stretch a lot, as it's a very common tool, by now. What do you think?
When I've been in studios that use only 44.1 kHz, I was really amazed of that choice, but thinking about it I found it was making sense, based on the type of clients they had, their workflow (one was recording through a 24 tracks tape recorder "to get the analog sound"), etc.
Thanks a lot for the show! 🙂
Thank you so much for this episode of "The Mastering Show" and for all these links! It is amazing and an invaluable input for pros as well as for enthusiasts and both of them!
Thank you - I'm a big fan!
Ulrich
Hi Ian,
Love your shows, but I must say I strongly disagree with anyone who says 16 bit is just as good as 24. Good 24 bit files CLEARLY have more sonic depth. I am wondering however that people who hear 16 and 24 just as good as each other must have significant hearing loss. Same perhaps, though not as sure, with sample rates. While technically we can't hear sounds above 20k, there are lots of things are ears perceive that aren't perceptible as "sound". For example each room we go into sounds different, even when there are no sounds actually being made. My theory, and I must say theory, is higher frequencies could be responsible for this.
Thank you for all your wonderful work on this show!!!!
Glad you like the show, but are you saying I have hearing damage ?!?
24-bit files clearly have more depth ONLY if dither wasn't used - and even then, only on some material.
Out of interest, I'm assuming you score 100% success on this test ?
http://www.audiocheck.net/blindtests_16vs8bit.php
Ian
Well I have since thought more about my perspective on this. It might not have so much to do with "hearing damage". It may have more to do with the type of recording. I have noticed that recordings that capture good amounts of spaciousness/depth in the sound, sound significantly better in 24 bit. I think a lot of the music that is being produced today has so little dynamic range and spaciousness, that recordings at 16/44.1 are perfectly suitable and actually in many cases sound even better than 24/96. But pure acoustic recordings with depth and nuance certainly sound better with higher "quality" files, especially when high quality converters are used.
Also I think the brain also makes psychoacoustic compensation on lower quality files sometimes. I noticed that if I listen to lots of "hi res" music, then go to compressed files, the compressed files will sound very much lacking in quality. But then after a while my brain adjusts and it doesn't sound so bad after a while. I think there is probably a psychoacoustic tendency of the brain to fill in the gap of the missing piece. Such as when the ear will hear low frequencies that aren't reproduced in a lower quality system because it hears the overtones and can fill in the gap accordingly.
I think a better test would be to play for someone a HD file for a long time. Perhaps someone sits and listens to a whole album of good quality acoustic music with a high level of nuance and depth. THEN play the lower res file immediately after. I think when just bouncing back and forth the brain has a hard time noticing differences because there is no time for the brain to adjust to the sounds of the individual recording. Also, as I said, loud rock/pop/electronic music there won't be much of a difference.
Sorry to butt in. When you say you hear a difference between 16/44 and 24/96, have you controlled all variables apart from the bit depth and sampling rate eg using a 24/96 file and a down-sampling/comparison program like Foobar with the DBX plug-in? If so, copy and paste your results as it would be intriguing.
Typically, when differences are heard, the files come from different masters, or not level matched or the listener knows which file is playing in advance of listening, thus bringing expectation biases and placebo effects leading to a perception of greater "spaciousness" etc etc.
I doubt anyone over 30 yo can hear over 20khz, and hence the difference between that and a 48khz upper range from a sample rate of 96 (assuming of course that there is actual music content at those frequencies). As for bit depth, the advantage of 24bits over 16 for production is well understood, in fact as Ian could attest, most studios these days use 32 or 48bits, but pretty well useless for playback.
The only difference between 24bits and 16 is an extra 48db of dynamic range, and nothing else. It is science and maths which gave us digital audio and there is nothing the science and maths doesn't understand about digital audio. There is no magic, though psychology can help us understand how perception can be influenced by beliefs - ie expectation biases and placebo effects.
Have a read of the opening post in the thread below. The guy who posted it is an engineer who studied under Bob Katz. He explains the 24bits vs 16 more succinctly than I can.
http://www.head-fi.org/t/415361/24bit-vs-16bit-the-myth-exploded
The most convincing reason I know for using reasonable bit rates of 44.1 kHz or 48 kHz is that most microphones do not capture anything above 20 kHz. Any ultrasonic content in recordings will likely be noise from analog gear. Of course there are mics (such as some of the Earthworks mics) that capture frequencies up to 30 or 40 kHz, so maybe if you have those mic it might be justified to use higher sampling rates. But honestly, how many home studio folks have those, or other mics that capture anything above 20 kHz? Even most pro studios' mics do not. I think most analog FX gear are rated only up to 20 kHz or maybe a bit higher. I really want to know, what ultrasonic content can actually be there that can contribute to the music?
What a superb show Ian! I absolutely love this Podcast. Despite having been a long term follower of your Production advice website and having done your Home Mastering Course, I still continue to learn and benefit from every topic you take on. If I'd heard this particular episode 2 years ago, I would have been clueless. The fact that I enjoyed the Geek talk and was able to follow, must mean I'm actually making progress. Keep up the good work !
While the benefits of 24bits (and higher) for recording, mixing and mastering are well understood, it is amazing that after 30 years we are still having debates of whether 24bits (in particular) or sample rates greater than 44.1 offer any benefit playback. Surely this alone raises a red flag because if there was a benefit, or at least a material improvement in sound quality, it would have been settled a long time ago. There also would be no need for hi res playback equipment to have "HD" indicators to feed into a placebo effect.
Many people believe that they can hear a difference on playback, but under properly controlled DBT the difference they hear always disappear or are found to be different masterings, a DAC performing better on one sample rate than another, and so on.
Even the meta study you quote, after accounting for all peer review over recent years (including the flawed ones) finds only a slight difference - and that is mainly in relation to MP3. In the audio engineering community we are still waiting on a "hi res" purveyor (step forward Neil Young) to fund a study like the comprehensive one in the link below which found that over 100 subjects (including trained musicians, music producers and audiophiles) in a period of over a year, at times using their own music, in their own home, on their own stereo in their own time could not pick the CD resolution from hi res at a rate greater than guessing.
The sad part about this is that it is another false hope, forcing the public to pay more for better mastered music (and even then it is not a guarantee) rather than doing right in the first place.
http://drewdaniels.com/audible.pdf